Error in the subroutine PSHDIR while running TRANSIENT simulation

MarcoC501 · June 22, 2021, 05:46

Hi,
I am trying to run the Time Transformation of the blade passages in the attached figure. I had some memory problems, so I am trying to run the "skeleton" version (no results data), just to see if the simulation is starting.
Anyway, I get a message error I have never seen in the Topology Simplification at the beginning of the simulation

Details of error:-
----------------
Error detected by routine PSHDIR
CDRNAM = /ABR/OUTPUT/ZN3 /PERIODC
CRESLT = NONE

Current Directory : /ABR/OUTPUT/ZN3

I am using a converged SS solution. Does someone know what is this problem linked to?

Thanks a lot

Opaque · June 22, 2021, 08:03

What version of the software are you running?

How many "transient rotor-stator" do you have in your setup? How many "frozen rotor/mixing plane" ones you got?

How are the different transformations defined?

There is something missing to determine the period of "ZN3". Recall ZN3 is an internal name in ANSYS CFX, but related to one of your domains.

MarcoC501 · June 22, 2021, 11:04

Hi Opaque,
thanks a lot as always.
I am running with CFX 2019 R3.
I have 2 transient stator-rotor at the interfaces between IGV-rotor and rotor-stator, but time transformation only between IGV-rotor (where I am more interested). There is also another error info when writing the recovery file

Details of error:-
----------------
Error detected by routine PSHDIR
CDRNAM = /ABR/OUTPUT/ZN3 /PERIODC
CRESLT = NONE

Current Directory : /ABR/OUTPUT/ZN3

+--------------------------------------------------------------------+
| Writing crash recovery file |
+--------------------------------------------------------------------+

Details of error:-
----------------
Error detected by routine PEEKR
CDANAM = ../ZN1 /BCP1 /AREA
CRESLT = NONE

Current Directory : /FLOW/MESH/TSTEP0/CLOOP0/ZN1

Opaque · June 22, 2021, 12:16

I would not worry about error messages after the message for writing the recovery file.

I recall seeing similar errors in the past (missing ZNn/PERIOD or ZNn/PERIODC).

1 - Do you have your Number of Passages in Component/360 set?
2 - Do you have an export surface for Harmonic Forced Response set?
3 - Have you tried setting up a 2nd Time Transformation on the other "transient rotor-stator" domain interface? Just select Option = None for both Side1/Side2 and see if the error persists.

MarcoC501 · June 22, 2021, 12:43

1- I set the number of passages (double checked just now)
2- I don't
3- If I try to set the time transformation just to rotor-stator, it doesn't work. If I set both (with the same passing period, number of timesteps, number of periods per run), it seems it is running.
Anyway, I am not fully sure of the reason why, and if this simulation is really going to work as I want.

Opaque · June 22, 2021, 14:31

Could you post the section under Transient Blade Row Models to see how you set up the Time Transformation?

MarcoC501 · June 22, 2021, 15:45

I set the time transformation IGV-rotor in the way shown in the pitcure. For the other time transformation, the setup is the same.
I can only check the residual trend from the solver manager (for the moment, because of memory problem), and so far it is not so significant.

Just an off topic question: while setting the monitors it was giving me an error that the "massFlowAve" (for total pressure, torque) function can be used only at point locators for time transformation discurbances. This means that I have to choose a point before and after the blade manually (maybe at 50% of the span)? Or from your experience is better to set it in a different way?
Thanks a lot!

Opaque · June 22, 2021, 16:42

Time transformation implies a different time in different nodes; therefore, any spatial accumulative operation is not valid.

For monitor points, there is no dependency on the neighbors; therefore, you can use them without issues.

You should add monitor points where you have knowledge of the expected signal, or in the region of interest while the solution progresses.

You can post-process the final results for overall studies.

On the time transformation settings, the 2nd time transformation you should use None, not Domain List nor Automatic. I think you meant to model one of the transient rotor-stator, not both correct?.

I strongly advise you to read the documentation for modeling using Time Transformation.

MarcoC501 · June 23, 2021, 04:33

If I set 'none' for the second time transformation, I get the same error in the routine PSHDIR. At least I am able to run the simulation by setting up 2 time transformation.
Anyway, I stopped the simulation because the residuals were going constant and increasing, which I do not like. I now introduced some monitors for the pressure at significant points.
Do you know how can I set the torque in this case? In the SS I simply set:
torque()@R1Blade
but now that I need to specify the cartesian coordinates I cannot do that anymore

June 22, 2021, 08:03		#2
Opaque Senior Member Join Date: Jun 2009 Posts: 1,819 Rep Power: 32	What version of the software are you running? How many "transient rotor-stator" do you have in your setup? How many "frozen rotor/mixing plane" ones you got? How are the different transformations defined? There is something missing to determine the period of "ZN3". Recall ZN3 is an internal name in ANSYS CFX, but related to one of your domains. __________________ Note: I do not answer CFD questions by PM. CFD questions should be posted on the forum.

June 22, 2021, 12:16		#4
Opaque Senior Member Join Date: Jun 2009 Posts: 1,819 Rep Power: 32	I would not worry about error messages after the message for writing the recovery file. I recall seeing similar errors in the past (missing ZNn/PERIOD or ZNn/PERIODC). 1 - Do you have your Number of Passages in Component/360 set? 2 - Do you have an export surface for Harmonic Forced Response set? 3 - Have you tried setting up a 2nd Time Transformation on the other "transient rotor-stator" domain interface? Just select Option = None for both Side1/Side2 and see if the error persists. __________________ Note: I do not answer CFD questions by PM. CFD questions should be posted on the forum.

June 22, 2021, 14:31		#6
Opaque Senior Member Join Date: Jun 2009 Posts: 1,819 Rep Power: 32	Could you post the section under Transient Blade Row Models to see how you set up the Time Transformation? __________________ Note: I do not answer CFD questions by PM. CFD questions should be posted on the forum.

June 22, 2021, 16:42		#8
Opaque Senior Member Join Date: Jun 2009 Posts: 1,819 Rep Power: 32	Time transformation implies a different time in different nodes; therefore, any spatial accumulative operation is not valid. For monitor points, there is no dependency on the neighbors; therefore, you can use them without issues. You should add monitor points where you have knowledge of the expected signal, or in the region of interest while the solution progresses. You can post-process the final results for overall studies. On the time transformation settings, the 2nd time transformation you should use None, not Domain List nor Automatic. I think you meant to model one of the transient rotor-stator, not both correct?. I strongly advise you to read the documentation for modeling using Time Transformation. __________________ Note: I do not answer CFD questions by PM. CFD questions should be posted on the forum.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Transient simulation -> Steady state solution	saisanthoshm88	CFX	13	April 10, 2024 04:43
a transient cfx simulation suddenly stopped writing .out and then .bak while running	mona.li	CFX	1	March 5, 2018 04:15
Transient simulation running better with larger timestep	Foxhunter	FLUENT	2	November 9, 2017 01:11
GUI crash and simulation engine still running	RPJones	FLOW-3D	2	November 9, 2010 08:18
transient simulation of a rotating rectangle	icesniffer	CFX	1	August 8, 2009 07:25

June 22, 2021, 11:04		#3
MarcoC501 Member MC Join Date: Apr 2021 Posts: 43 Rep Power: 5	Hi Opaque, thanks a lot as always. I am running with CFX 2019 R3. I have 2 transient stator-rotor at the interfaces between IGV-rotor and rotor-stator, but time transformation only between IGV-rotor (where I am more interested). There is also another error info when writing the recovery file Details of error:- ---------------- Error detected by routine PSHDIR CDRNAM = /ABR/OUTPUT/ZN3 /PERIODC CRESLT = NONE Current Directory : /ABR/OUTPUT/ZN3 +--------------------------------------------------------------------+ \| Writing crash recovery file \| +--------------------------------------------------------------------+ Details of error:- ---------------- Error detected by routine PEEKR CDANAM = ../ZN1 /BCP1 /AREA CRESLT = NONE Current Directory : /FLOW/MESH/TSTEP0/CLOOP0/ZN1

June 22, 2021, 12:43		#5
MarcoC501 Member MC Join Date: Apr 2021 Posts: 43 Rep Power: 5	1- I set the number of passages (double checked just now) 2- I don't 3- If I try to set the time transformation just to rotor-stator, it doesn't work. If I set both (with the same passing period, number of timesteps, number of periods per run), it seems it is running. Anyway, I am not fully sure of the reason why, and if this simulation is really going to work as I want.

June 23, 2021, 04:33		#9
MarcoC501 Member MC Join Date: Apr 2021 Posts: 43 Rep Power: 5	If I set 'none' for the second time transformation, I get the same error in the routine PSHDIR. At least I am able to run the simulation by setting up 2 time transformation. Anyway, I stopped the simulation because the residuals were going constant and increasing, which I do not like. I now introduced some monitors for the pressure at significant points. Do you know how can I set the torque in this case? In the SS I simply set: torque()@R1Blade but now that I need to specify the cartesian coordinates I cannot do that anymore